Study On Predicting Model Of Tectonic Deformed Coal Thickness Based On Regression Random Forest

Coal and gas outburst is one of the great threats in the process of coal mining in China.The research shows that the existence of tectonic deformed coal will promote the occurrence of coal and gas outburst,and the tectonic deformed coal development area is often the prone area of gas outburst.Therefore,if the gas outburst area can be identified by predicting the thickness of tectonic deformed coal,it can provide a theoretical basis for the prevention and control of coal and gas outburst.Considering that the single objective method is often used to predict the thickness of tectonic deformed coal at present,and the internal relationship of different types of tectonic deformed coal is not used,this paper constructs a regression random forest model to synchronously predict the thickness of various tectonic deformed coal.Firstly,the original coal seam attribute data is dimensionally reduced by principal component analysis,and independent principal components are obtained to represent the original attributes.secondly,considering that the performance of the regression random forest is not as good as its performance in classification prediction,the secondary training process is introduced to give the individual model and feature attribute weights,so as to improve the prediction effect of the individual model and improve the integration efficiency.At the same time,considering the weak global exploration capability of the whale optimization algorithm and the waste of search capability resources,the overall exploration capability of the algorithm is improved by changing the ratio of the exploration and mining capabilities of the whale optimization algorithm.And change the search method of the search agent in exploration and spiral hunting and expand the parameter range to improve the hunting ability of the whale optimization algorithm and obtain better prediction results.In this paper,the model is applied to the actual coal seam attribute data of Luling coal mine,three types of tectonic deformed coal are predicted synchronously,and the thickness of three types of tectonic deformed coal is predicted.By comparing the model with single objective prediction model and multi-target support vector regression model,it is concluded that the model has better prediction performance.Therefore,the prediction model proposed in this paper has good prediction effect,and can be applied to the synchronous prediction of multiple tectonic deformed coal in other gas outburst prone areas.